https://bugs.kde.org/show_bug.cgi?id=413317
--- Comment #5 from wolthera <griffinval...@gmail.com> --- So... bisecting says that this is the bad commit: 5ccac066cd8d98e655b7d14349dc011205ac096f is the first bad commit commit 5ccac066cd8d98e655b7d14349dc011205ac096f Author: Dmitry Kazakov <dimul...@gmail.com> Date: Mon Sep 16 21:14:41 2019 +0300 Fix wrong borders on EdgeDetection and HeightToNormalMap filters The color data should be premultiplied by the alpha channel to get correct result on layers with transparency. BUG:411922 libs/image/kis_edge_detection_kernel.cpp | 16 ++++++++++++++++ libs/pigment/KoColorSpace.h | 5 +++++ libs/pigment/KoColorSpaceAbstract.h | 5 +++++ .../kis_convert_height_to_normal_map_filter.cpp | 3 --- .../filters/edgedetection/kis_edge_detection_filter.cpp | 3 --- 5 files changed, 26 insertions(+), 6 deletions(-) (https://invent.kde.org/graphics/krita/-/commit/5ccac066cd8d98e655b7d14349dc011205ac096f) I guess premultiplying is what leads to this bug??? -- You are receiving this mail because: You are watching all bug changes.